private double GetDotProduct(DocumentBase d1, DocumentBase d2)
{
if (cache.ContainsKey(d1) && cache[d1].ContainsKey(d2))
{
return(cache[d1][d2]);
}
if (cache.ContainsKey(d2) && cache[d2].ContainsKey(d1))
{
return(cache[d2][d1]);
}
var weights1 = _tfIdfWeights[d1];
var weights2 = _tfIdfWeights[d2];
double sum = 0;
foreach (var term in weights1.Keys)
{
var w1 = weights1[term];
var w2 = weights2[term];
sum += w1 * w2;
}
if (!cache.ContainsKey(d1))
{
cache[d1] = new Dictionary <DocumentBase, double>();
}
if (!cache[d1].ContainsKey(d2))
{
cache[d1][d2] = sum;
}
return(sum);
}
private double GetMagnitude(DocumentBase document)
{
if (!magnitudes.ContainsKey(document))
{
var weights = _tfIdfWeights[document];
var magnitude = weights.Values.Sum(weight => weight * weight);
magnitudes[document] = Math.Sqrt(magnitude);
}
return(magnitudes[document]);
}
public Dictionary <string, double> GetWeights(DocumentBase document)
{
return(_tfIdfWeights.ContainsKey(document) ? _tfIdfWeights[document] : null);
}